]>
Commit | Line | Data |
---|---|---|
1 | #include "cache.h" | |
2 | #include "grep.h" | |
3 | #include "xdiff-interface.h" | |
4 | ||
5 | void append_header_grep_pattern(struct grep_opt *opt, enum grep_header_field field, const char *pat) | |
6 | { | |
7 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
8 | p->pattern = pat; | |
9 | p->origin = "header"; | |
10 | p->no = 0; | |
11 | p->token = GREP_PATTERN_HEAD; | |
12 | p->field = field; | |
13 | *opt->pattern_tail = p; | |
14 | opt->pattern_tail = &p->next; | |
15 | p->next = NULL; | |
16 | } | |
17 | ||
18 | void append_grep_pattern(struct grep_opt *opt, const char *pat, | |
19 | const char *origin, int no, enum grep_pat_token t) | |
20 | { | |
21 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
22 | p->pattern = pat; | |
23 | p->origin = origin; | |
24 | p->no = no; | |
25 | p->token = t; | |
26 | *opt->pattern_tail = p; | |
27 | opt->pattern_tail = &p->next; | |
28 | p->next = NULL; | |
29 | } | |
30 | ||
31 | static int isregexspecial(int c) | |
32 | { | |
33 | return c == '\0' || is_glob_special(c) || | |
34 | c == '$' || c == '(' || c == ')' || c == '+' || | |
35 | c == '.' || c == '^' || c == '{' || c == '|'; | |
36 | } | |
37 | ||
38 | static int is_fixed(const char *s) | |
39 | { | |
40 | while (!isregexspecial(*s)) | |
41 | s++; | |
42 | return !*s; | |
43 | } | |
44 | ||
45 | static void compile_regexp(struct grep_pat *p, struct grep_opt *opt) | |
46 | { | |
47 | int err; | |
48 | ||
49 | if (opt->fixed || is_fixed(p->pattern)) | |
50 | p->fixed = 1; | |
51 | if (opt->regflags & REG_ICASE) | |
52 | p->fixed = 0; | |
53 | if (p->fixed) | |
54 | return; | |
55 | ||
56 | err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
57 | if (err) { | |
58 | char errbuf[1024]; | |
59 | char where[1024]; | |
60 | if (p->no) | |
61 | sprintf(where, "In '%s' at %d, ", | |
62 | p->origin, p->no); | |
63 | else if (p->origin) | |
64 | sprintf(where, "%s, ", p->origin); | |
65 | else | |
66 | where[0] = 0; | |
67 | regerror(err, &p->regexp, errbuf, 1024); | |
68 | regfree(&p->regexp); | |
69 | die("%s'%s': %s", where, p->pattern, errbuf); | |
70 | } | |
71 | } | |
72 | ||
73 | static struct grep_expr *compile_pattern_or(struct grep_pat **); | |
74 | static struct grep_expr *compile_pattern_atom(struct grep_pat **list) | |
75 | { | |
76 | struct grep_pat *p; | |
77 | struct grep_expr *x; | |
78 | ||
79 | p = *list; | |
80 | switch (p->token) { | |
81 | case GREP_PATTERN: /* atom */ | |
82 | case GREP_PATTERN_HEAD: | |
83 | case GREP_PATTERN_BODY: | |
84 | x = xcalloc(1, sizeof (struct grep_expr)); | |
85 | x->node = GREP_NODE_ATOM; | |
86 | x->u.atom = p; | |
87 | *list = p->next; | |
88 | return x; | |
89 | case GREP_OPEN_PAREN: | |
90 | *list = p->next; | |
91 | x = compile_pattern_or(list); | |
92 | if (!x) | |
93 | return NULL; | |
94 | if (!*list || (*list)->token != GREP_CLOSE_PAREN) | |
95 | die("unmatched parenthesis"); | |
96 | *list = (*list)->next; | |
97 | return x; | |
98 | default: | |
99 | return NULL; | |
100 | } | |
101 | } | |
102 | ||
103 | static struct grep_expr *compile_pattern_not(struct grep_pat **list) | |
104 | { | |
105 | struct grep_pat *p; | |
106 | struct grep_expr *x; | |
107 | ||
108 | p = *list; | |
109 | switch (p->token) { | |
110 | case GREP_NOT: | |
111 | if (!p->next) | |
112 | die("--not not followed by pattern expression"); | |
113 | *list = p->next; | |
114 | x = xcalloc(1, sizeof (struct grep_expr)); | |
115 | x->node = GREP_NODE_NOT; | |
116 | x->u.unary = compile_pattern_not(list); | |
117 | if (!x->u.unary) | |
118 | die("--not followed by non pattern expression"); | |
119 | return x; | |
120 | default: | |
121 | return compile_pattern_atom(list); | |
122 | } | |
123 | } | |
124 | ||
125 | static struct grep_expr *compile_pattern_and(struct grep_pat **list) | |
126 | { | |
127 | struct grep_pat *p; | |
128 | struct grep_expr *x, *y, *z; | |
129 | ||
130 | x = compile_pattern_not(list); | |
131 | p = *list; | |
132 | if (p && p->token == GREP_AND) { | |
133 | if (!p->next) | |
134 | die("--and not followed by pattern expression"); | |
135 | *list = p->next; | |
136 | y = compile_pattern_and(list); | |
137 | if (!y) | |
138 | die("--and not followed by pattern expression"); | |
139 | z = xcalloc(1, sizeof (struct grep_expr)); | |
140 | z->node = GREP_NODE_AND; | |
141 | z->u.binary.left = x; | |
142 | z->u.binary.right = y; | |
143 | return z; | |
144 | } | |
145 | return x; | |
146 | } | |
147 | ||
148 | static struct grep_expr *compile_pattern_or(struct grep_pat **list) | |
149 | { | |
150 | struct grep_pat *p; | |
151 | struct grep_expr *x, *y, *z; | |
152 | ||
153 | x = compile_pattern_and(list); | |
154 | p = *list; | |
155 | if (x && p && p->token != GREP_CLOSE_PAREN) { | |
156 | y = compile_pattern_or(list); | |
157 | if (!y) | |
158 | die("not a pattern expression %s", p->pattern); | |
159 | z = xcalloc(1, sizeof (struct grep_expr)); | |
160 | z->node = GREP_NODE_OR; | |
161 | z->u.binary.left = x; | |
162 | z->u.binary.right = y; | |
163 | return z; | |
164 | } | |
165 | return x; | |
166 | } | |
167 | ||
168 | static struct grep_expr *compile_pattern_expr(struct grep_pat **list) | |
169 | { | |
170 | return compile_pattern_or(list); | |
171 | } | |
172 | ||
173 | void compile_grep_patterns(struct grep_opt *opt) | |
174 | { | |
175 | struct grep_pat *p; | |
176 | ||
177 | if (opt->all_match) | |
178 | opt->extended = 1; | |
179 | ||
180 | for (p = opt->pattern_list; p; p = p->next) { | |
181 | switch (p->token) { | |
182 | case GREP_PATTERN: /* atom */ | |
183 | case GREP_PATTERN_HEAD: | |
184 | case GREP_PATTERN_BODY: | |
185 | compile_regexp(p, opt); | |
186 | break; | |
187 | default: | |
188 | opt->extended = 1; | |
189 | break; | |
190 | } | |
191 | } | |
192 | ||
193 | if (!opt->extended) | |
194 | return; | |
195 | ||
196 | /* Then bundle them up in an expression. | |
197 | * A classic recursive descent parser would do. | |
198 | */ | |
199 | p = opt->pattern_list; | |
200 | opt->pattern_expression = compile_pattern_expr(&p); | |
201 | if (p) | |
202 | die("incomplete pattern expression: %s", p->pattern); | |
203 | } | |
204 | ||
205 | static void free_pattern_expr(struct grep_expr *x) | |
206 | { | |
207 | switch (x->node) { | |
208 | case GREP_NODE_ATOM: | |
209 | break; | |
210 | case GREP_NODE_NOT: | |
211 | free_pattern_expr(x->u.unary); | |
212 | break; | |
213 | case GREP_NODE_AND: | |
214 | case GREP_NODE_OR: | |
215 | free_pattern_expr(x->u.binary.left); | |
216 | free_pattern_expr(x->u.binary.right); | |
217 | break; | |
218 | } | |
219 | free(x); | |
220 | } | |
221 | ||
222 | void free_grep_patterns(struct grep_opt *opt) | |
223 | { | |
224 | struct grep_pat *p, *n; | |
225 | ||
226 | for (p = opt->pattern_list; p; p = n) { | |
227 | n = p->next; | |
228 | switch (p->token) { | |
229 | case GREP_PATTERN: /* atom */ | |
230 | case GREP_PATTERN_HEAD: | |
231 | case GREP_PATTERN_BODY: | |
232 | regfree(&p->regexp); | |
233 | break; | |
234 | default: | |
235 | break; | |
236 | } | |
237 | free(p); | |
238 | } | |
239 | ||
240 | if (!opt->extended) | |
241 | return; | |
242 | free_pattern_expr(opt->pattern_expression); | |
243 | } | |
244 | ||
245 | static char *end_of_line(char *cp, unsigned long *left) | |
246 | { | |
247 | unsigned long l = *left; | |
248 | while (l && *cp != '\n') { | |
249 | l--; | |
250 | cp++; | |
251 | } | |
252 | *left = l; | |
253 | return cp; | |
254 | } | |
255 | ||
256 | static int word_char(char ch) | |
257 | { | |
258 | return isalnum(ch) || ch == '_'; | |
259 | } | |
260 | ||
261 | static void show_line(struct grep_opt *opt, const char *bol, const char *eol, | |
262 | const char *name, unsigned lno, char sign) | |
263 | { | |
264 | if (opt->null_following_name) | |
265 | sign = '\0'; | |
266 | if (opt->pathname) | |
267 | printf("%s%c", name, sign); | |
268 | if (opt->linenum) | |
269 | printf("%d%c", lno, sign); | |
270 | printf("%.*s\n", (int)(eol-bol), bol); | |
271 | } | |
272 | ||
273 | static void show_name(struct grep_opt *opt, const char *name) | |
274 | { | |
275 | printf("%s%c", name, opt->null_following_name ? '\0' : '\n'); | |
276 | } | |
277 | ||
278 | static int fixmatch(const char *pattern, char *line, regmatch_t *match) | |
279 | { | |
280 | char *hit = strstr(line, pattern); | |
281 | if (!hit) { | |
282 | match->rm_so = match->rm_eo = -1; | |
283 | return REG_NOMATCH; | |
284 | } | |
285 | else { | |
286 | match->rm_so = hit - line; | |
287 | match->rm_eo = match->rm_so + strlen(pattern); | |
288 | return 0; | |
289 | } | |
290 | } | |
291 | ||
292 | static int strip_timestamp(char *bol, char **eol_p) | |
293 | { | |
294 | char *eol = *eol_p; | |
295 | int ch; | |
296 | ||
297 | while (bol < --eol) { | |
298 | if (*eol != '>') | |
299 | continue; | |
300 | *eol_p = ++eol; | |
301 | ch = *eol; | |
302 | *eol = '\0'; | |
303 | return ch; | |
304 | } | |
305 | return 0; | |
306 | } | |
307 | ||
308 | static struct { | |
309 | const char *field; | |
310 | size_t len; | |
311 | } header_field[] = { | |
312 | { "author ", 7 }, | |
313 | { "committer ", 10 }, | |
314 | }; | |
315 | ||
316 | static int match_one_pattern(struct grep_opt *opt, struct grep_pat *p, char *bol, char *eol, enum grep_context ctx) | |
317 | { | |
318 | int hit = 0; | |
319 | int saved_ch = 0; | |
320 | regmatch_t pmatch[10]; | |
321 | ||
322 | if ((p->token != GREP_PATTERN) && | |
323 | ((p->token == GREP_PATTERN_HEAD) != (ctx == GREP_CONTEXT_HEAD))) | |
324 | return 0; | |
325 | ||
326 | if (p->token == GREP_PATTERN_HEAD) { | |
327 | const char *field; | |
328 | size_t len; | |
329 | assert(p->field < ARRAY_SIZE(header_field)); | |
330 | field = header_field[p->field].field; | |
331 | len = header_field[p->field].len; | |
332 | if (strncmp(bol, field, len)) | |
333 | return 0; | |
334 | bol += len; | |
335 | saved_ch = strip_timestamp(bol, &eol); | |
336 | } | |
337 | ||
338 | again: | |
339 | if (!p->fixed) { | |
340 | regex_t *exp = &p->regexp; | |
341 | hit = !regexec(exp, bol, ARRAY_SIZE(pmatch), | |
342 | pmatch, 0); | |
343 | } | |
344 | else { | |
345 | hit = !fixmatch(p->pattern, bol, pmatch); | |
346 | } | |
347 | ||
348 | if (hit && opt->word_regexp) { | |
349 | if ((pmatch[0].rm_so < 0) || | |
350 | (eol - bol) <= pmatch[0].rm_so || | |
351 | (pmatch[0].rm_eo < 0) || | |
352 | (eol - bol) < pmatch[0].rm_eo) | |
353 | die("regexp returned nonsense"); | |
354 | ||
355 | /* Match beginning must be either beginning of the | |
356 | * line, or at word boundary (i.e. the last char must | |
357 | * not be a word char). Similarly, match end must be | |
358 | * either end of the line, or at word boundary | |
359 | * (i.e. the next char must not be a word char). | |
360 | */ | |
361 | if ( ((pmatch[0].rm_so == 0) || | |
362 | !word_char(bol[pmatch[0].rm_so-1])) && | |
363 | ((pmatch[0].rm_eo == (eol-bol)) || | |
364 | !word_char(bol[pmatch[0].rm_eo])) ) | |
365 | ; | |
366 | else | |
367 | hit = 0; | |
368 | ||
369 | if (!hit && pmatch[0].rm_so + bol + 1 < eol) { | |
370 | /* There could be more than one match on the | |
371 | * line, and the first match might not be | |
372 | * strict word match. But later ones could be! | |
373 | * Forward to the next possible start, i.e. the | |
374 | * next position following a non-word char. | |
375 | */ | |
376 | bol = pmatch[0].rm_so + bol + 1; | |
377 | while (word_char(bol[-1]) && bol < eol) | |
378 | bol++; | |
379 | if (bol < eol) | |
380 | goto again; | |
381 | } | |
382 | } | |
383 | if (p->token == GREP_PATTERN_HEAD && saved_ch) | |
384 | *eol = saved_ch; | |
385 | return hit; | |
386 | } | |
387 | ||
388 | static int match_expr_eval(struct grep_opt *o, | |
389 | struct grep_expr *x, | |
390 | char *bol, char *eol, | |
391 | enum grep_context ctx, | |
392 | int collect_hits) | |
393 | { | |
394 | int h = 0; | |
395 | ||
396 | switch (x->node) { | |
397 | case GREP_NODE_ATOM: | |
398 | h = match_one_pattern(o, x->u.atom, bol, eol, ctx); | |
399 | break; | |
400 | case GREP_NODE_NOT: | |
401 | h = !match_expr_eval(o, x->u.unary, bol, eol, ctx, 0); | |
402 | break; | |
403 | case GREP_NODE_AND: | |
404 | if (!collect_hits) | |
405 | return (match_expr_eval(o, x->u.binary.left, | |
406 | bol, eol, ctx, 0) && | |
407 | match_expr_eval(o, x->u.binary.right, | |
408 | bol, eol, ctx, 0)); | |
409 | h = match_expr_eval(o, x->u.binary.left, bol, eol, ctx, 0); | |
410 | h &= match_expr_eval(o, x->u.binary.right, bol, eol, ctx, 0); | |
411 | break; | |
412 | case GREP_NODE_OR: | |
413 | if (!collect_hits) | |
414 | return (match_expr_eval(o, x->u.binary.left, | |
415 | bol, eol, ctx, 0) || | |
416 | match_expr_eval(o, x->u.binary.right, | |
417 | bol, eol, ctx, 0)); | |
418 | h = match_expr_eval(o, x->u.binary.left, bol, eol, ctx, 0); | |
419 | x->u.binary.left->hit |= h; | |
420 | h |= match_expr_eval(o, x->u.binary.right, bol, eol, ctx, 1); | |
421 | break; | |
422 | default: | |
423 | die("Unexpected node type (internal error) %d", x->node); | |
424 | } | |
425 | if (collect_hits) | |
426 | x->hit |= h; | |
427 | return h; | |
428 | } | |
429 | ||
430 | static int match_expr(struct grep_opt *opt, char *bol, char *eol, | |
431 | enum grep_context ctx, int collect_hits) | |
432 | { | |
433 | struct grep_expr *x = opt->pattern_expression; | |
434 | return match_expr_eval(opt, x, bol, eol, ctx, collect_hits); | |
435 | } | |
436 | ||
437 | static int match_line(struct grep_opt *opt, char *bol, char *eol, | |
438 | enum grep_context ctx, int collect_hits) | |
439 | { | |
440 | struct grep_pat *p; | |
441 | if (opt->extended) | |
442 | return match_expr(opt, bol, eol, ctx, collect_hits); | |
443 | ||
444 | /* we do not call with collect_hits without being extended */ | |
445 | for (p = opt->pattern_list; p; p = p->next) { | |
446 | if (match_one_pattern(opt, p, bol, eol, ctx)) | |
447 | return 1; | |
448 | } | |
449 | return 0; | |
450 | } | |
451 | ||
452 | static int grep_buffer_1(struct grep_opt *opt, const char *name, | |
453 | char *buf, unsigned long size, int collect_hits) | |
454 | { | |
455 | char *bol = buf; | |
456 | unsigned long left = size; | |
457 | unsigned lno = 1; | |
458 | struct pre_context_line { | |
459 | char *bol; | |
460 | char *eol; | |
461 | } *prev = NULL, *pcl; | |
462 | unsigned last_hit = 0; | |
463 | unsigned last_shown = 0; | |
464 | int binary_match_only = 0; | |
465 | const char *hunk_mark = ""; | |
466 | unsigned count = 0; | |
467 | enum grep_context ctx = GREP_CONTEXT_HEAD; | |
468 | ||
469 | if (buffer_is_binary(buf, size)) { | |
470 | switch (opt->binary) { | |
471 | case GREP_BINARY_DEFAULT: | |
472 | binary_match_only = 1; | |
473 | break; | |
474 | case GREP_BINARY_NOMATCH: | |
475 | return 0; /* Assume unmatch */ | |
476 | break; | |
477 | default: | |
478 | break; | |
479 | } | |
480 | } | |
481 | ||
482 | if (opt->pre_context) | |
483 | prev = xcalloc(opt->pre_context, sizeof(*prev)); | |
484 | if (opt->pre_context || opt->post_context) | |
485 | hunk_mark = "--\n"; | |
486 | ||
487 | while (left) { | |
488 | char *eol, ch; | |
489 | int hit; | |
490 | ||
491 | eol = end_of_line(bol, &left); | |
492 | ch = *eol; | |
493 | *eol = 0; | |
494 | ||
495 | if ((ctx == GREP_CONTEXT_HEAD) && (eol == bol)) | |
496 | ctx = GREP_CONTEXT_BODY; | |
497 | ||
498 | hit = match_line(opt, bol, eol, ctx, collect_hits); | |
499 | *eol = ch; | |
500 | ||
501 | if (collect_hits) | |
502 | goto next_line; | |
503 | ||
504 | /* "grep -v -e foo -e bla" should list lines | |
505 | * that do not have either, so inversion should | |
506 | * be done outside. | |
507 | */ | |
508 | if (opt->invert) | |
509 | hit = !hit; | |
510 | if (opt->unmatch_name_only) { | |
511 | if (hit) | |
512 | return 0; | |
513 | goto next_line; | |
514 | } | |
515 | if (hit) { | |
516 | count++; | |
517 | if (opt->status_only) | |
518 | return 1; | |
519 | if (binary_match_only) { | |
520 | printf("Binary file %s matches\n", name); | |
521 | return 1; | |
522 | } | |
523 | if (opt->name_only) { | |
524 | show_name(opt, name); | |
525 | return 1; | |
526 | } | |
527 | /* Hit at this line. If we haven't shown the | |
528 | * pre-context lines, we would need to show them. | |
529 | * When asked to do "count", this still show | |
530 | * the context which is nonsense, but the user | |
531 | * deserves to get that ;-). | |
532 | */ | |
533 | if (opt->pre_context) { | |
534 | unsigned from; | |
535 | if (opt->pre_context < lno) | |
536 | from = lno - opt->pre_context; | |
537 | else | |
538 | from = 1; | |
539 | if (from <= last_shown) | |
540 | from = last_shown + 1; | |
541 | if (last_shown && from != last_shown + 1) | |
542 | fputs(hunk_mark, stdout); | |
543 | while (from < lno) { | |
544 | pcl = &prev[lno-from-1]; | |
545 | show_line(opt, pcl->bol, pcl->eol, | |
546 | name, from, '-'); | |
547 | from++; | |
548 | } | |
549 | last_shown = lno-1; | |
550 | } | |
551 | if (last_shown && lno != last_shown + 1) | |
552 | fputs(hunk_mark, stdout); | |
553 | if (!opt->count) | |
554 | show_line(opt, bol, eol, name, lno, ':'); | |
555 | last_shown = last_hit = lno; | |
556 | } | |
557 | else if (last_hit && | |
558 | lno <= last_hit + opt->post_context) { | |
559 | /* If the last hit is within the post context, | |
560 | * we need to show this line. | |
561 | */ | |
562 | if (last_shown && lno != last_shown + 1) | |
563 | fputs(hunk_mark, stdout); | |
564 | show_line(opt, bol, eol, name, lno, '-'); | |
565 | last_shown = lno; | |
566 | } | |
567 | if (opt->pre_context) { | |
568 | memmove(prev+1, prev, | |
569 | (opt->pre_context-1) * sizeof(*prev)); | |
570 | prev->bol = bol; | |
571 | prev->eol = eol; | |
572 | } | |
573 | ||
574 | next_line: | |
575 | bol = eol + 1; | |
576 | if (!left) | |
577 | break; | |
578 | left--; | |
579 | lno++; | |
580 | } | |
581 | ||
582 | free(prev); | |
583 | if (collect_hits) | |
584 | return 0; | |
585 | ||
586 | if (opt->status_only) | |
587 | return 0; | |
588 | if (opt->unmatch_name_only) { | |
589 | /* We did not see any hit, so we want to show this */ | |
590 | show_name(opt, name); | |
591 | return 1; | |
592 | } | |
593 | ||
594 | /* NEEDSWORK: | |
595 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
596 | * which feels mostly useless but sometimes useful. Maybe | |
597 | * make it another option? For now suppress them. | |
598 | */ | |
599 | if (opt->count && count) | |
600 | printf("%s%c%u\n", name, | |
601 | opt->null_following_name ? '\0' : ':', count); | |
602 | return !!last_hit; | |
603 | } | |
604 | ||
605 | static void clr_hit_marker(struct grep_expr *x) | |
606 | { | |
607 | /* All-hit markers are meaningful only at the very top level | |
608 | * OR node. | |
609 | */ | |
610 | while (1) { | |
611 | x->hit = 0; | |
612 | if (x->node != GREP_NODE_OR) | |
613 | return; | |
614 | x->u.binary.left->hit = 0; | |
615 | x = x->u.binary.right; | |
616 | } | |
617 | } | |
618 | ||
619 | static int chk_hit_marker(struct grep_expr *x) | |
620 | { | |
621 | /* Top level nodes have hit markers. See if they all are hits */ | |
622 | while (1) { | |
623 | if (x->node != GREP_NODE_OR) | |
624 | return x->hit; | |
625 | if (!x->u.binary.left->hit) | |
626 | return 0; | |
627 | x = x->u.binary.right; | |
628 | } | |
629 | } | |
630 | ||
631 | int grep_buffer(struct grep_opt *opt, const char *name, char *buf, unsigned long size) | |
632 | { | |
633 | /* | |
634 | * we do not have to do the two-pass grep when we do not check | |
635 | * buffer-wide "all-match". | |
636 | */ | |
637 | if (!opt->all_match) | |
638 | return grep_buffer_1(opt, name, buf, size, 0); | |
639 | ||
640 | /* Otherwise the toplevel "or" terms hit a bit differently. | |
641 | * We first clear hit markers from them. | |
642 | */ | |
643 | clr_hit_marker(opt->pattern_expression); | |
644 | grep_buffer_1(opt, name, buf, size, 1); | |
645 | ||
646 | if (!chk_hit_marker(opt->pattern_expression)) | |
647 | return 0; | |
648 | ||
649 | return grep_buffer_1(opt, name, buf, size, 0); | |
650 | } |